U.S. patent number 7,466,690 [Application Number 10/490,568] was granted by the patent office on 2008-12-16 for traffic restriction for a network with qos transmission.
This patent grant is currently assigned to Siemens Aktiengesellschaft. Invention is credited to Karl Schrodi.
United States Patent |
7,466,690 |
Schrodi |
December 16, 2008 |
Traffic restriction for a network with QoS transmission
Abstract
The invention relates to a method, a boundary node and a server
for restricting the traffic in a packet-oriented, connectionless
network for an efficient QoS transmission of prioritized data
packets. According to the invention, reliability checks are carried
out that include a reliability check with respect to the network
input and the network output. The reliability checks allow to check
whether resources meeting the requirements to transmission of a
group of data packets of a priority class are available in the
network. The invention allows to avoid resource shortages,
especially at the network input and network output, thereby
safeguarding QoS transmission.
Inventors: |
Schrodi; Karl (Geretsried,
DE) |
Assignee: |
Siemens Aktiengesellschaft
(Munich, DE)
|
Family
ID: |
26010189 |
Appl.
No.: |
10/490,568 |
Filed: |
September 20, 2002 |
PCT
Filed: |
September 20, 2002 |
PCT No.: |
PCT/DE02/03538 |
371(c)(1),(2),(4) Date: |
March 19, 2004 |
PCT
Pub. No.: |
WO03/026229 |
PCT
Pub. Date: |
March 27, 2003 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20040264376 A1 |
Dec 30, 2004 |
|
Foreign Application Priority Data
|
|
|
|
|
Sep 20, 2001 [DE] |
|
|
101 46 349 |
Dec 14, 2001 [DE] |
|
|
101 61 546 |
|
Current U.S.
Class: |
370/352; 370/360;
370/395.21; 370/395.3; 370/400; 709/225; 709/238 |
Current CPC
Class: |
H04L
12/5601 (20130101); H04L 45/00 (20130101); H04L
45/302 (20130101); H04L 47/10 (20130101); H04L
47/125 (20130101); H04L 47/2408 (20130101); H04L
47/2433 (20130101); H04L 47/32 (20130101); H04L
47/781 (20130101); H04L 47/801 (20130101); H04L
47/805 (20130101); H04L 47/822 (20130101); H04L
47/70 (20130101); H04L 2012/562 (20130101); H04L
2012/5632 (20130101); H04L 2012/5651 (20130101) |
Current International
Class: |
H04L
12/66 (20060101) |
Field of
Search: |
;370/235-237,254-255,351-360,396,398,395.2,395.21,395.32,395.42,395.43,395.5,395.52,400-401
;709/224-225,238 |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
100 14 522 |
|
Oct 2001 |
|
DE |
|
1 133 112 |
|
Sep 2001 |
|
EP |
|
WO 93/11622 |
|
Jun 1993 |
|
WO |
|
WO 99/66676 |
|
Dec 1999 |
|
WO |
|
WO 01/54448 |
|
Jul 2001 |
|
WO |
|
Other References
H Hussmann et al: "An Edge Device for Supporting Internet
Integrated Services Over Switched ATM Networks", Sep. 1998,
Interworking Conference, Toronto, CANADA, pp. 6-7. cited by other
.
Blake, S. Black, D., Carlson, M. [u.a.]: "An Architecture for
Differentiated Service", no date. cited by other .
Nichols, K. Jocobson, V., Zhang, L.; A Two-Bit Differentiated
Services Architecture for the Internet, no date. cited by
other.
|
Primary Examiner: Tieu; Binh K.
Attorney, Agent or Firm: Bell, Boyd & Lloyd
Claims
The invention claimed is:
1. A method for restricting the traffic in a packet-oriented
connectionless network, comprising: receiving a voice over internet
protocol (VoIP) call by a first access network from a first
terminal, the access network communicatively coupled to an IP
network having a plurality of nodes and to a first control unit;
sending a connection setup from the first access network to the
first control unit; transmitting a connection setup from the first
control unit to a second control unit, the second control unit
communicatively coupled a second access network that is
communicatively coupled to the IP network and to a second terminal;
identifying from the plurality of nodes an ingress node, the
identification by the first control unit; transmitting requirements
for the VoIP call from the first control unit to the ingress node;
checking by the ingress mode that the respective node fulfills the
requirements; identifying from the plurality of nodes an egress
node, the identification by the second control unit; transmitting
requirements for the VoIP call from the second control unit to the
egress node; checking by the egress mode that the respective node
fulfills the requirements; and establishing the bearer path from
the first terminal to the second terminal over the IP network so
that a group of data packets flow between the first terminal and
the ingress node via the first access network and flow between the
second terminal and the egress node via the second access
network.
2. The method according to claim 1, wherein the group of data
packets is specified by the data packets of a traffic stream or by
the data packets of the priority class aggregated at a port.
3. The method according to claim 2, wherein the traffic stream is
specified by a flow, the traffic stream allocated to a connection
or including data packets with common address information.
4. The method according to claim 1, wherein a framework of the
reliability checks of a rate of capacity utilization within an
entire capacity of the network and an available bandwidth, the
input-side and output-side network accesses are used when
transmitting the group, of data packets.
5. The method according to claim 4, wherein the traffic of the
priority class of the traffic stream and/or the traffic of a
priority class transmitted is higher or the same as the priority
class of the group of data packets.
6. The method according to claim 1, wherein the group of data
packets is rejected if there is a negative result of the check by
the egress node or by the ingress node.
7. The method according to claim 1, wherein traffic parameters for
the group of data packets routed through the network are reported
in the network including information concerning required
transmission resources.
8. The method according to claim 7, wherein the transmission
resources are reserved in accordance with the traffic
parameters.
9. The method according to claim 7, wherein a check is performed
during transmission of the traffic to determine whether or not
reported traffic parameters are adhered to.
10. The method according to claim 7, wherein the non-reported data
packets are blocked-out or discarded.
11. The method according to claim 7, wherein the reliability checks
are only carried out for prioritized traffic.
12. The method according to claim 1, wherein the non-prioritized
data packets are transmitted with an enhanced quality.
Description
CROSS REFERENCE OF RELATED APPLICATIONS
This application is the U.S. National Stage of International
Application No. PCT/DE02/03538, filed Sep. 20, 2002 and claims the
benefit thereof. The International Application claims the benefits
of German application No. 10146349.9 DE filed Sep. 20, 2001, and
German application No. 10161546.9 DE filed Dec. 14, 2001, both of
the applications are incorporated by reference herein in their
entirety.
FIELD OF INVENTION
The invention relates to a method, a network, a boundary node and a
server for restricting the traffic in a packet-oriented,
connectionless network for an efficient QoS transmission of
prioritized data packets.
BACKGROUND OF INVENTION
The integration of telecommunications and data services leads to a
series of requirements for the switching technology and networks.
In order to be suitable for infotainment and commercial traffic at
the same time, the networks used should have a high capacity, allow
transmission in realtime, be reliable and guarantee a high degree
of safety. In addition to this, a further condition is to keep
costs as low as possible.
Previously, data services were to a large extent processed via IP
networks (networks based on the Internet Protocol) that operate
packet-oriented and connectionless at the IP level. Progress in
router technology have led to the development of IP routers that,
with a view to the capacity of the switched data traffic and delay
times because of queues, basically allow telecommunications
services and bandwidth-intensive services such as video-on-demand
or video conferences in real time.
Difficulties are encountered in the case of a high rate of capacity
utilization of the IP network for which the delay times increase
exponentially or the excessive aggregation of data traffic on
individual routes which then act as a bottleneck and limit the
transmission rate.
Because of these difficulties exist, a high service quality
generally referred to in the literature as quality-of-service (QoS)
cannot be guaranteed to the desired extent for conventional IP
networks.
Further developments aim at obtaining better information about the
service quality without impinging on the lack of complexity and
flexibility of the IP network.
The differentiated services (diff-derv) model is based on the
observation that the best-effort handling of data packets in the IP
network leads to the above-mentioned difficulties in guaranteeing
service quality. Conventionally, data packets are transferred
quickly and completely without guaranteeing the reliability and
safety of the transmission. During high utilization or overloading
of the network the service quality is impaired by delays or
discarding of data packets.
The diff-serv concept aims at improving the service quality for
services with high quality requirements by introducing service
classes. In this context, the term CoS (class of service) model is
often mentioned. The numbers 2474 and 2475 describe the diff-serv
concept published by the IETF. The RFCs 2638 and 2998 deal with
further aspects of the concept. Within the framework of the
diff-serv concept, a DS (differentiated services) field in the IP
header of the data packets prioritizes the packet traffic by
setting the DSCP (DS code point) parameters. This prioritization
takes place with a "per hop" resource allocation, i.e. the packets
are treated differently for the nodes depending on the class of
service specified by the DSCP parameters in the DS field. The
expression per-hop behavior (PBH) is used in this context. For
example, in the course of a PBH, priority is given to higher
classes of service with regard to the arrangement and processing of
queues in the case of nodes.
The central elements of a network based on the diff-serv concept
are the DS subnetworks--in English often called the DS domain or
single routing domain--and the DS boundary nodes. In many cases, a
subnetwork conforms to the network of a service provider (service
provider domain). For the DS boundary nodes a distinction is made
between the DS ingress nodes and the DS egress nodes. Data packets
reach a DS subnetwork via a DS ingress node and leave the
subnetwork via a DS egress node. Therefore, a DS boundary node can
then unite the functionality of a DS ingress node for incoming
traffic and a DS egress node for outgoing traffic. The
functionality of the DS boundary node includes selecting data
packets according to the DSCP parameters and marking data packets
by means of DSCP parameters. In addition, by means of devices for
managing and conditioning the traffic, it is also possible to carry
out control measures such as measuring data flows, distributing
data packets to queues or rejecting data packets in the DS
subnetwork. These control measures are often carried out in the DS
ingress node or the DS egress node. Within the framework of traffic
conditioning, data packets can be classified and compared to a
traffic profile (e.g. bandwidth, resources) provided for the
corresponding class of service. In the case of deviations from the
traffic profile, measures such as arranging a queue or rejecting
data packets can be carried out.
Typically, flows or connections are prioritized in a DS subnetwork.
Prioritization takes place in a DS ingress node for the
corresponding DS subnetwork, if required, by setting or changing
the DSCP parameters. The core nodes of the DSCP parameters are read
and assigned to the resources according to the prioritization. The
DSCP parameters and resource assignment of the individual core
nodes are interpreted separately from one another (per-hop
behavior). For the egress nodes, changing the DSCP parameters is
possibly canceled, i.e. the DSCP is reset to the original value. In
this way, the DSCP can be adapted locally, i.e. depending on the
subnetwork and its features.
The diff-serv concept avoids complex reservation procedures of
routes or bandwidth and prioritizes the data traffic. When
transmitting via several subnetworks, classes of service are
specified by means of so-called service level agreements (SLA) for
the entire transmission and converted by the individual subnetworks
as described above within the framework of traffic conditioning.
However, in practice temporary and/or local shortages, for example,
by the aggregation of data traffic to individual routes also
occurs. Usually, data packets with the same destination from the
time they came together in a node, follow the same set route. For
shortages the diff-serv concept sees to it that data packets with a
low priority are first of all delayed or rejected. Therefore, the
transmission quality for the high-priority data packets is
improved, but quality standards e.g. for real-time transmission
cannot be guaranteed. Within the framework of the diff-serv concept
it would only be possible to guarantee transmission with QoS, i.e.
a transmission for which specific quality statements are given and
adhered to if the traffic profiles are adjusted with such a low
utilization of the subnetwork that load peaks were catered for by
reserve bandwidths. This is not usually undertaken for reasons of
cost, i.e. because of the resulting low network utilization. For
this reason, within the context of the diff-serv concept, reference
is made to a CoS (class of service) approach rather than a QoS
(quality of service) approach.
SUMMARY OF INVENTION
The object of the invention is to allow an efficient transmission
of data packets with QoS for packet-switched, connectionless
networks.
The invention aims at an efficient QoS-capable network with
packet-switched and connection-oriented operation. The concept for
the network includes the following three considerations: A high
efficiency requires flexibility in selecting the routes of data
packets or their distribution. For example resource shortages can
be avoided by dynamically changing or adapting routes. A high
complexity is avoided while local decisions are made (e.g. via
routes, queues, rejecting data packets, etc.), e.g. in each case by
the corresponding router. Non-local states or reservations of
routes are avoided. Decisions about routes can be made depending on
the data packet or depending on the connection or the flow. Thus,
the distribution of individual data packets is the most flexible.
Guaranteeing QoS requires stringent boundaries and monitoring with
a view to utilizing the network.
This patent application deals with the third aspect.
According to the invention, reliability checks allow checks to be
made on whether or not the resources meeting the requirements for
transmitting a group of data packets of a priority class are
available in the corresponding network. According to the invention,
reliability checks are carried out for at least one priority class
for groups of data packets to be routed via the network. These
reliability checks include a reliability check with respect to the
network input and network output that can also be carried out at
the network input and network output. Only in the case of data
packet groups for which all the reliability checks are positive,
transmission with the priority class of the data packets is
allowed. On the other hand, for groups of data packets for which
one of the reliability checks is negative, a different procedure is
carried out (claim 1).
Therefore, one group of data packets can be specified by the data
packets of a traffic stream or by the data packets aggregated at a
physical port, e.g. at a network access (claim 2). A traffic
stream, for example, corresponds to a flow, includes the data
packets of a connection or includes data packets with identical
address information such as all data packets with the same source
or destination (claim 3).
In order to guarantee a QoS transmission, overloading the total
capacity of the network and resource shortages must be avoided. In
this sense, in the case of reliability checks, the group of data
packets to be transmitted can be evaluated according to parameters
such as average data and/or packet rate, peak rate, etc. and checks
can be made on whether or not sufficient transmission capacity is
available in order to transmit traffic streams with the required
service quality. In addition, the reliability checks also ensure
that sufficient resources (e.g. bandwidth, queue capacity, etc.)
are available both at the input-side and output-side (claim 4). The
check can depend on the priority class of the data packets or on
the traffic with the same or higher priority (claim 5). For
example, without a reliability check of the output-side, the
resources of several inputs of the network could be transmitted to
the same output and in this way result in a shortage at the output.
Therefore, quality guarantees could then not be adhered to, but at
best only a quality undertaking with respect to the prioritized
handling of traffic streams in the same way as for the diff-serv
concept.
For example, as a criterion for a positive result of the
reliability checks, a threshold value can be used that is
determined depending on the capacity of the network input and the
network output used in each case, the total capacity of the network
and the desired quality or priority class, etc. For example,
traffic parameters such as the average data and/or the packet rate
and the peak rate are reported for a group of data packets to be
transmitted with a priority (claim 7). The desired priority class
can also be reported. Alternatively, the priority class is
determined on the basis of parameters or requirements such as the
maximum loss rate and real-time transmission.
However, it is also conceivable that for each priority class there
are several threshold values on the basis of different evaluation
parameters that must all be kept separately or in the corresponding
dependencies to one another. For a negative result of the
reliability checks, the transmission of the group of data packets
can be rejected (claim 6) or the transmission can be carried out
with a low priority or not prioritized.
On the basis of the reliability checks, the resources (e.g.
bandwidth, queues) can be reserved in accordance with the required
quality features (claim 8). This reservation is usually the
associated network access and output as well as the entire network
load (e.g. capacity, handling in queues according to the
prioritization).
Adhering to the central traffic parameters reported for the
relevant traffic streams such as the transmission rate should
possibly be monitored in order to guarantee adhering to the limit
values or the threshold values for the utilization of the network
(claim 9). The monitoring function--the terms traffic enforcement
and policing are often used for this--will compare in a suitable
way the traffic parameters specified when requesting the quality
features with the actual traffic parameters of the corresponding
traffic stream. Non-reported data packets can be blocked out (claim
10). Known traffic shaping mechanisms such as leaky bucket or token
bucket can also be used at the network input. Possible overload
prevention measures are as follows: Rejecting data packets Marking
data packets Buffering data packets Switching over or blocking the
traffic stream Converting the data packets violating the agreement
or the entire relevant data stream to a lower priority class or
handling according to the best effort approach
A QoS transmission of traffic streams with a corresponding priority
class or traffic class requires a corresponding handling of the
priority class. It is useful to only utilize a part of the entire
capacity with prioritized traffic. The other part of the network
capacity is then utilized with non-prioritized traffic. This
non-prioritized traffic can then be treated according to the
best-effort principle (claim 12). The reliability checks can then
be restricted to the prioritized traffic (claim 11). This
restriction of the prioritized traffic ensures that the capacity of
the network can be fully utilized without the load peaks having a
negative effect on the prioritized traffic. The quality level by
means of which non-prioritized traffic is transmitted then
virtually acts as a buffer for the prioritized traffic. A possible
procedure of setting boundaries for the utilization with
prioritized traffic is that a fixed maximum percentage utilization
for traffic with the same or higher priority class must be given
for each priority class. For example, it could be possible to set
in a network with two priority classes for traffic with the higher
traffic class the limit to a utilization with 30% and to specify
the limit for traffic with the higher or lower priority class to
60%. For non-prioritized traffic, a minimum capacity of 40% would
then remain.
When transmitting data packets, priority class-specific quality
features could then be guaranteed in such a way that they allow a
QoS transmission. Therefore, the concept according to the invention
assumes that a suitable quality of service (QoS) is oriented at the
specific service. For example, the human sense organs can to a
certain extent process incomplete information without it resulting
in any subjective quality losses. For the interactive control of
machines (e.g. remote control of robots), the requirements are much
clearer according to the circumstances. Criteria that are
correspondingly more stringent should then be used. Therefore,
depending on the service, criteria or limits can then be defined
that guarantee a QoS transmission. For packet-oriented transmission
these criteria are as follows: Type and extent of possible
information losses Fixed and/or variable delays The temporal
consistence (sequence) of the information.
By adhering to utilization limits for the network, possibly within
the framework of a service-specifically pre-defined statistical
probability, and restricting the prioritized traffic as well as by
good traffic distribution and limiting the traffic at the accesses
and outputs of the network or the physical ports, statistical
values for the quality loss factors can be specified. QoS services
can be guaranteed with the aid of these statistical values and
their variance.
BRIEF DESCRIPTION OF DRAWINGS
Two variants of the object according to the invention are shown
below within the framework of the embodiment. They are as
follows:
FIG. 1: System with data transmission via a network according to
the invention
FIG. 2: A network according to the invention
FIG. 3: Diagram of different routes for routing two flows in one
network according to the invention
FIG. 4: Schematic diagram of different routes for routing two flows
with the same destination in one network according to the
invention.
DETAILED DESCRIPTION OF DRAWING
For the sake of clarity it is assumed that the invention is used
within the framework of a telephone call made via an IP network
IPN, i.e. voice over IP (VoIP). This IP network IPN can for example
be a single routing domain of the Internet. Telephone calls are
subject to real-time requirements. Therefore, the relevant data
traffic is prioritized. Accordingly, the object according to the
invention can be used analogously for all the other services for
which a prioritization of the data traffic is needed. Examples of
such services are video-on-demand, Web conferences, multimedia
applications, etc.
FIG. 1 shows a system with VoIP transmission. Via access networks
AN-A and AN-B (AN: for access network), the telecommunications
terminals TLN-A and TLN-B are connected to a public network that
includes the IP network IPN. Within the framework of the two
variants of the embodiment it is assumed that terminal TLN-A sets
up a connection to terminal TLN-B by means of a telephone call. In
this case a distinction is made between the service level SL (SL:
for service level) and the network level (NL: for network level)
that is shown in the figure by means of a dotted line. At the
service level SL, signaling SIG(VA,DS) (SIG(VA,DS) for: signaling
the connection setup and service control) of the connection setup
and the service control takes place. For this purpose, control
units CCP-A and CCP-B (CCP: for call control point) e.g. media
gateway controller or switching devices are connected to the access
networks AN-A and AN-B of terminals TLN-A and TLN-B. Useful data is
transmitted to the network level NL and is at least sometimes
routed to network IPN (IPN: for IP network) according to the
invention. The network IPN operates packet-oriented and
connectionless. Useful data packets that are transferred within the
framework of the telephone call from terminal TNL-A to terminal
TLN-B reach the network IPN via the boundary node IgNd (IgNd: for
ingress node) and leave it again via the boundary node EgNd (EgNd:
for egress node).
In the course of the connection setup at the service level SL, the
call request is signaled from terminal TLN-A via the access network
AN-A to the control unit CCP-A. Terminal TLN-A is authorized, for
example, at the basis of name or address information. The called
terminal TLN-B or the allocated control unit CCP-B is then
identified and localized. Usually, a connection setup message is
transmitted from the control unit CCP-A to the control unit CCP-B.
Relevant information is extracted in the control unit CCP-B and at
the access network AC-B the availability of terminal TLN-B is
checked and relevant information is requested. The connection setup
message is then acknowledged by the control unit CCP-B, sent to the
control unit CCP-A and information such as address information of
terminal TLN-B required for the connection is transmitted. The
connection setup at the service level SL can then be concluded. In
the case of a successful connection setup, useful data can then be
exchanged at the network level NL.
In the case of the selected example, voice information in
real-time, i.e. with QoS is exchanged as useful data. For QoS
transmission, reliability checks are carried out according to the
invention. Signaling takes place within the framework of these
reliability checks, e.g. when transmitting the desired quality
requirements, when transmitting the result of the reliability
checks, etc. This signaling is designated as QoS signaling below. A
distinction is made between two variants depending on whether or
not QoS signaling is carried out at the service level SL or the
network level NL within the framework of the reliability
checks.
For QoS signaling at the service level SL the method according to
the invention can be as follows: via the control units of the
service level, CCP-Ig and CCP-Eg, the boundary nodes IgNd and EgNd
are identified via which the useful data is transmitted as useful
data packets. The control units CCP-Ig and CCP-Eg shown in FIG. 2
can be but do not have to be identical to the control units CCP-A
and CC-P allocated to the access networks AN-A and AN-B.
Usually, many control units have to be involved in the signaling
for long-distance calls at the service level SL. These control
units then have direct access to only one section or a sub-section
of the entire transmission route for the useful data. The control
units CCP-Ig (CCP-Ig: for call control point at ingress node) and
CCP-Eg (CCP-Eg: for call control point at egress node) are
identified in such a way that they communicate with the boundary
nodes IgNd and EgNd. Therefore, the two control units CCP-Ig and
CCP-Eg can also coincide with each other.
FIG. 2 shows further boundary nodes. The core nodes CoNd (CoNd: for
core node) are not shown individually. A control server NCS (NCS:
for network control server) is also shown by means of which the
network management tasks are observed.
According to the invention, the reliability checks NAC (NAC: for
network admission control) regarding the network input and the
network output are performed e.g. in the boundary nodes IgNd or
EgNd. Via the control units CCP-Ig and CCP-Eg of service level SL,
the requirements for the VoIP telephone call are transmitted to the
boundary nodes IgNd and EgNd between terminals TLN-A and TLN-B. The
requirements can include in addition to the relevant traffic
parameters such as the bandwidth and the QoS requirements
additional parameters regarding reliability, safety, etc. The
result of the reliability checks at the network boundaries is
signaled to the control units CCP-A and CCP-B. Depending on the
results of the reliability checks, the transmission of useful data
is either released or blocked on the A side and possibly an
alternative route is searched for, for the purpose of useful data
transmission (in English also known as bearer redirection). The
reliability checks can also take place during the connection setup
at the service level SL. For reliability checks during the
connection setup, the connection setup can possibly be aborted for
a negative result.
In the case of QoS signaling at the network level NL, the
reliability checks are only initiated after the connection setup at
the service level SL. After the successful connection setup at the
service level SL, the QoS signaling is released at the network
level NL and the relevant information such as the B-side address
information is transmitted. This can take place by a corresponding
message from the control unit CCP-A to a device positioned in the
access network AC-A, e.g. media gateway. For QoS signaling it is
also possible to make available from the service level SL, e.g.
from the control unit CCP-A, a program structure, e.g. a Java
Applet of the unit in the access network. By means of the address
information requested within the framework of the service setup at
the service level and, if required, by means of the downloaded
program structure, a signaling message is sent at the network level
to the B-side terminal TLN-B or the B-side access network AN-B. Via
these and possibly other signaling messages, the boundary nodes
IgNd and EgNd are localized and the reliability checks initiated in
the boundary nodes. The results of the reliability checks are then
signaled by messages at the network level NL to the A-side access
network AN-A or the A-side subscriber TLN-A.
In addition to the reliability checks regarding the network input
and output, a reliability check concerning the entire capacity of
the network is carried out. This reliability check can e.g. also
take place in one of the boundary nodes IgNd or EgNd, distributed
to both boundary nodes IgNd and EgNd, or in a server provided for
this in the network.
FIG. 3 is a diagram of different routes for routing two flows in
one network according to the invention. The limit of the network is
shown by means of a dotted line. Network nodes are shown by means
of circles in which case the circles represent the rank nodes that
intersect the dotted line. Arrows show possible routes for a flow.
The dotted arrows are possible routes for a flow that reach the
network for the rank node C and are transmitted to the rank node D
where the data packets of the flow again leave the network. Drawn
arrows show possible routes for a flow that are routes from the
boundary node. A to the boundary node B. For most core nodes there
is more than one alternative or branching for the route of the
flow. In the case of a specific flow, the different alternatively
possible route sections from a core node to the next node, i.e. for
the "next hop" are designated below as branching compartments of
the flow for the corresponding core node. For core nodes that lie
on possible routes of both flows, it is indicated whether or not
the branching compartments of the flow are identical i, sometimes
disjunctive t or disjunctive d.
Possible routes are determined from the parameters of the network
such as topology, capacity or the individual route sections, delay
times, etc. Decisions about the route compartments via which a
packet or a group of packets have to be transmitted, is made
locally depending on the traffic parameters applicable at that
moment. In this way, a relatively uniform utilization of the
network is reached and load shortages are avoided.
FIG. 4 differs from FIG. 3 because of the fact that both flows
leave the network in the case of boundary node B. The inner core
that possibly lies on the routes of both flows, the distribution
compartments of both flows are identical in which case the nodes
are marked with an i. The distribution compartments could also
deviate from one another within the framework of traffic control or
traffic shaping. However, they will at least to a large extent be
identical for flows with the same destination. The reliability
checks according to the invention bring about that the boundary
nodes B and the core nodes that are topologically adjacent to the
boundary node are not overloaded by the arriving data packets of
both flows. For example, if both flows cannot be transmitted with
the reported parameters, one of them will not be permitted. When
restricting the reliability checks to the network input, the
configuration in FIG. 4 could lead to a shortage at boundary node B
thereby endangering the guarantee undertaking of the QoS.
* * * * *